Features

Grammar

Connect a database or SharePoint site and build the knowledge graph on a visual canvas.

• ✓Auto-detect tables, columns, and foreign-key relationships
• ✓Edit aliases, readings, derived fields, and enumerations
• ✓Save workspaces and revisit as your schema evolves

Analyst

Agentic harness for validating the graph: plans query paths, runs SQL, cites facts, and refuses when the graph does not carry the answer.

• ✓Multi-step reasoning with tool use against live data
• ✓Scoped investigations with citations and findings
• ✓Execution history and per-query audit

Dashboard

Live KPI cards and charts with built-in plain-English Q&A. Ask in the slide-over panel, pin answers to the grid.

• ✓Plain-English questions return charts or tables
• ✓Shareable chat configs; session persists while the panel is open
• ✓Resize presets and drag-handle reorder; portals scoped per viewer
• ✓Live refresh on the underlying data; works on desktop, tablet, and phone

Lightweight graphs per domain

Build a focused graph for a specific use case instead of modelling the whole database at once. A finance graph and a warehouse graph live side by side.

Multiple graphs per dashboard

A single dashboard can mix widgets backed by different graphs and different connections. No one-graph-fits-all compromise.

Dashboards grow from questions

Every widget starts as a question in Dashboard’s chat panel that an analyst chose to pin. Dashboards accumulate from real questions, not pre-built reports.

The graph is the deliverable

On the Builder tier the graph exports as portable JSON. Version-control it, embed it, or keep it as an artifact when an engagement ends.

Entities and value types

Tables become entities; columns become typed values. Optional namespace qualifier for schema-prefixed tables.

Fact types with readings

Binary relationships expressed as natural-language predicates, with explicit forward and inverse readings.

Reference schemes

Preferred identifier per entity. The column used in JOIN and WHERE clauses.

Data types

Text, Number, Date, Boolean, Currency. Chooses the correct SQL functions and aggregation rules.

Enumerated value sets

Closed-world domains on value or entity types; produces precise WHERE ... IN (...) clauses.

Alternate readings

Extra phrasings of the same fact for synonym coverage when users ask the question differently.

Sample values

Concrete example rows attached to entities for disambiguation and query grounding.

Role names

Handles multi-reference joins (e.g. DepartureAirport vs ArrivalAirport) and self-referential relationships.

Derived value types

Row-level formulas such as Revenue - COGS, including cross-entity references like Quantity * Product.UnitPrice.

Aliases

Human-readable names for cryptic legacy columns and tables; physical names stay intact for SQL.

Uniqueness and mandatory roles

Functional dependencies and existence constraints declared per role; drives cardinality and join behaviour.

Alternate keys

Additional unique attributes beyond the reference scheme (for example, email address on a Person).

Natural language to SQL

Questions are resolved against the graph, not the raw schema. If the semantics don’t match, the query doesn’t run.

Agentic graph probe

Analyst runs multi-step investigations against saved dataset configs, with SQL execution and structured findings.

Read-only enforcement

DML and DDL keywords (INSERT, UPDATE, DELETE, DROP, ALTER…) are blocked at the API layer before execution.

Execution limits

Per-query 30-second timeout and 10,000-row cap. Free-tier users have a monthly AI call limit.

Audit log

Every query records user, timestamp, duration, rows returned, and error message if any.

Charts and tables

Dashboard’s chat panel renders results as charts, KPI cards, or tables. Any chart can be pinned to the dashboard you’re viewing.

Teams and portals

Invite by email, assign Analyst or Strategist, and grant individual chat configs or dashboard portals per member. Strategists execute queries through the owner’s saved connection. No password sharing.

Graph export and API

Builder tier exports the graph as portable JSON. Per-user ormle_… API keys, passed in the X-Api-Key header, authorize calls to the /gramapi surface for your own pipelines.

Read-only SQL

Write operations blocked server-side before the database ever sees them.

Encrypted credentials

Database passwords encrypted at rest. Never sent to the browser, never placed in AI prompts.

LLM isolation

The model sees the graph only. It never receives the raw schema, connection strings, or secrets.

Two-factor authentication

TOTP MFA available on every account; pairs with any authenticator app.

Rate limiting

Per-user throttles on queries and AI calls protect the database and the account.

Audit and CSRF

Full audit log of query execution; CSRF tokens enforced on every authenticated JSON endpoint.